[runejoha]

Hi, I know this is a little absurd collation but lets take a look at the Mhz speed the G4 performs in contrast to AMD or Intel.

I will say the price of the laptops and names of the laptops here to make sure that the differences of the whole systemes alltogheter is "similar", not that the price make that big difference.

The "best" (depens of the usage of course) processor from Apple (or Motorola, but you get the picture is 1.33 GHz which is in a laptop, in this case the Powebook 17'', price $2,999.

The "best" processor from for example HP Compaq is Intel� Pentium� M processor 1.70GHz ( hp compaq business notebook nc8000 price $2,849.001 )

There are a lot of potential differeneces in these systems to justify this differencer, but let us try to look at some of the most central and critical facts for this subject.

What makes a processor fast:

1.The instruction count.
2.The clock cycle time
3.clock cycles per instruction

The instruction count is larger for a RISC architecture, which means this is a potential bottleneck for the G4. The clock cycle time is as we can see much faster for the "best" mobile Pentium processor for a laptop (1.70 GHz) at the same level of the PowerBook than the G4 1.33 GHz. This means that the clock cycles per instruction has to be very low in the G4 to "beat" the Pentium M. I presume that the instruction count, which is larger for the RISCs, can be disregarded in the comparison because the CISC intruction can take a lot of clock cycles. This leaves, if I am not very wrong with my assumtions, only the clock cyle time left, and the Pentium M has a better performance.

So, even though you always can argument for that the processor performance never is the bottelenck in a system anyway, and this is not really an important subject, that the PCs have, in the same price levels, better processors? This means that maybe Apple should change to Intel ptrocessors, at least for their laptops (it seems like they have some problems put the G5 intop the Powerbook for at least a year or so, and the same arguments also goes for the G5 vs the new AMD ..)

Other drawbacks I know about:

The Apple computers have less resolution than the PCs in average. The arguments to justify this is becasue of standardization, that is is better to read on a aplle screen etc. but is it not always to have the possibility to have good resolution?

Another drawback is that some programs are not released on the OS X, but this is for niche endusers, and are in the big picture not a real problem. The Apple computers are beacuse of this not a temting consume for example gamers. I even wonder if an applemachine ever can run Doom3 etc.

As an end comment I hope you do not take this as an offend. I am curious because I am intrested in the Apple computers myself, and I ask the questions to learn more about this, for me, new area.

I have never seen a better OS than Panther.

regards
runejoha

[hardcat1970]

our company just bought a 1.5GHZ M IBM ThinkPad T40, and it runs very fast and it looks pretty good compared to other windows laptop.

It weights about 4.5lbs same as my 12" powerbook and it is thin and has a 14" screen.
But the thing i like about is that you can easily replace its ram and hard drive. I also like the build in battery, it is on the back, its small and you can put an extra battery on the drive bay for 9+ hours operation.

Why can't apple build something like this?
The only thing i don't like about this thinkpad is it's running on windows xp.

[Fellow2000]

Apple does not need to or want to change to INTEL. Steve himself says they are very happy with what IBM is doing right now.

Far as the speeds, you have to remember both machines are using 2 different OS's. For instance, I know someone with an old blueberry g3 that runs much faster than his pentium 4 machine because windows degrades with use.

It's like having a corvette engine and a corolla engine, but if the corvette engine is on square wheels, while the corolla is on on round ones, which do you think is going to go faster?

Apple simply does not need to compete with the raw speeds of processors because no matter how slow the proc, the OS handles everything in a much better way.

[runejoha]

>>Apple does not need to or want to change to INTEL. Steve himself says they are very happy with what IBM is doing right now.<<

Probably because of economical reasons?


>>It's like having a corvette engine and a corolla engine, but if the corvette engine is on square wheels, while the corolla is on on round ones, which do you think is going to go faster?<<

Is that so? I have always heard that Windows runs faster than OS X, one reasons because of their Graphical API (in contrast to the win32 which is very fast, and much better than xwindows), but that Panther now is just as fast as windows. And another ting, you can run Linux or NetBSD etc. on a x86 arch, as well on a Apple, so this is not a discussion about the OSes but the system performance!

But, what you DO forgett is that a computer is more than its OS. A CPU is needed for convertions (MP3 etc.) or for demanding games or programs, and then you might need the raw power from Intel. So why is a Powerbook so expensive in contrast to the Compaq which apparently perform better?

Regards
runejoha

[Link]

Because the compaq doesn't perform better.

You yourself might have pointed this out. While apple typically uses the most powerful chipset they can get for their processors, that compaq will have whatever they chose to give it the biggest profit margin (except in this case it's limited to intel chipsets.. hrm.. maybe the rest of the board is crap?)

Either way I wouldn't reccomend anything less then a thinkpad or alienware machine (I'd have to look at their motherboard though) in comparison to the powerbook: THE GOOD centrino notebooks tend to be equal to or more expensive then powerbooks.

OS X uses the OpenGL graphic subsystem, MS uses DirectX/Directdraw for their graphic subsystem, which has 2 speeds: slow and stable, and fast and crappy. OpenGL is slowly becoming the standard (and even used in longhorn if I'm correct) as it's cross-platform compatible with pretty much any OS.

While the PPCG4 is pretty much now in a ditch performance wise I wouldn't worry: The g4 was not IBM's fault, the g5 was and that is a rather nice processor. The portable versions of this CPU are still off in the future but I'm sure apple has plans whatever they may be.

[OMGWTFBBQ]

Originally posted by Link:

Either way I wouldn't reccomend anything less then a thinkpad or alienware machine (I'd have to look at their motherboard though) in comparison to the powerbook: THE GOOD centrino notebooks tend to be equal to or more expensive then powerbooks.

Thinkpads are a love 'em or hate 'em deal. Many people (usually Linux lovers) think they are the best. Others hate them, yet have to use them for their jobs.
I personally love Linux, but hate the Thinkpads. They are tough and tanklike - but I can't stand the little button navigation thing - I want a trackpad personally - and don't tell me to get a mouse - for what I do, I use the laptop as a laptop, nothing else attached - otherwise I'd just use a desktop.

As for Alienware - all they do is brand the Sager laptops with their own shells and logos. They don't do anything special to the hardware other than perhaps make up a buzzword so that they can charge more.
I used to love Alienware - then I bought one of their systems, had massive amounts of problems, and I will now never buy from them again, nor will any company I own or consult for, nor will I recommend them to anyone.
That said - Sager makes solid laptops - they are what nearly all of the HP/Compaq laptops are really underneath the graphic branding.

As for the original question of a G4 compared to an Intel/AMD chip - I can go into their cache levels and what tasks they are likely to be better at...
Or I could just say that I run a benchmark on any new system that I am exposed to so that I know how fast it is and what speed-up or slow-down I can expect when I move my code to my clustered machines.

The script is in Perl, which is nice because a Perl script will run on any OS that has Perl ported to it... which most definitely includes FreeBSD, Linux, Windows, and Mac - so that means it runs on all of the OSes that I work with.
The script is designed to measure the speed of the processor - not RAM - it is a small script.
That said, it isn't so terribly small that different cache sizes wouldn't help/hurt performance.

On my WinXP Home laptop that has a 1Ghz Athlon 4 in it - which is their old version of their mobile processor (they have a new version which I think they call the Athlon XP M, but I could be mistaken on that) - the script took 2 seconds to run.
On my webserver or the equivalent of a single node of one of my clusters, it is an Athlon XP 2200+ running on either FreeBSD or Linux - they both ran the test is 1 second.
My new AlBook with a 1.25Ghz G4 in it ran the test in 4 seconds.

For what I do, I don't really need my laptop to be fast at running my code. My code will run for a long time, so unless my laptop is incredibly fast, I will likely need to put it to sleep and interrupt whatever process is running at least twice in the day.
I just need a laptop that is durable and ergonomic - it is icing on the cake if it looks good too.

Ideally one would have a processor built specifically for your task at hand - but since we can't all afford to have things fabricated for us - you should look for the processor that meets your needs. If you need XYZ, then get XYZ.
If what you do doesn't really matter if it is done in 2 seconds or 4 seconds, then get the one that has the ancillary benefits that you need/want.

that is my take on it - Macs are pricey, so only go for them if you need the OS and/or user experience - otherwise, Intel/AMD will nearly always give you more bang for the buck if that is what you are after.

[hardcat1970]

Originally posted by OMGWTFBBQ:
Thinkpads are a love 'em or hate 'em deal. Many people (usually Linux lovers) think they are the best. Others hate them, yet have to use them for their jobs.
I personally love Linux, but hate the Thinkpads. They are tough and tanklike - but I can't stand the little button navigation thing - I want a trackpad personally - and don't tell me to get a mouse - for what I do, I use the laptop as a laptop, nothing else attached - otherwise I'd just use a desktop.

As for Alienware - all they do is brand the Sager laptops with their own shells and logos. They don't do anything special to the hardware other than perhaps make up a buzzword so that they can charge more.
I used to love Alienware - then I bought one of their systems, had massive amounts of problems, and I will now never buy from them again, nor will any company I own or consult for, nor will I recommend them to anyone.
That said - Sager makes solid laptops - they are what nearly all of the HP/Compaq laptops are really underneath the graphic branding.

As for the original question of a G4 compared to an Intel/AMD chip - I can go into their cache levels and what tasks they are likely to be better at...
Or I could just say that I run a benchmark on any new system that I am exposed to so that I know how fast it is and what speed-up or slow-down I can expect when I move my code to my clustered machines.

The script is in Perl, which is nice because a Perl script will run on any OS that has Perl ported to it... which most definitely includes FreeBSD, Linux, Windows, and Mac - so that means it runs on all of the OSes that I work with.
The script is designed to measure the speed of the processor - not RAM - it is a small script.
That said, it isn't so terribly small that different cache sizes wouldn't help/hurt performance.

On my WinXP Home laptop that has a 1Ghz Athlon 4 in it - which is their old version of their mobile processor (they have a new version which I think they call the Athlon XP M, but I could be mistaken on that) - the script took 2 seconds to run.
On my webserver or the equivalent of a single node of one of my clusters, it is an Athlon XP 2200+ running on either FreeBSD or Linux - they both ran the test is 1 second.
My new AlBook with a 1.25Ghz G4 in it ran the test in 4 seconds.

For what I do, I don't really need my laptop to be fast at running my code. My code will run for a long time, so unless my laptop is incredibly fast, I will likely need to put it to sleep and interrupt whatever process is running at least twice in the day.
I just need a laptop that is durable and ergonomic - it is icing on the cake if it looks good too.

Ideally one would have a processor built specifically for your task at hand - but since we can't all afford to have things fabricated for us - you should look for the processor that meets your needs. If you need XYZ, then get XYZ.
If what you do doesn't really matter if it is done in 2 seconds or 4 seconds, then get the one that has the ancillary benefits that you need/want.

that is my take on it - Macs are pricey, so only go for them if you need the OS and/or user experience - otherwise, Intel/AMD will nearly always give you more bang for the buck if that is what you are after.

actually, the t40 has a little red button and a trackpad, that's what makes it so good. After using the thinkpad for couple of hours, i feel very comfortable with the trackpad and the soft trackpad button. As a matter of fact, it is one of the best trackpad i have ever use beside the pismo. Of course, the alu powerbook keyboard is the best.

[nagromme]

When price comparing to a PowerBook, don't just look at the "first glance" features like MHz and HD size. PowerBooks offer a LOT of extra functionality that's not present on many Windows laptops:

Gigabit ethernet

Ethernet that auto-corrects between crossover and patch cables (no need to care which is which!)

Three kinds of wireless (a, g, BT)

Lit keys

DVI and VGA and S-Video and composite out

Video spanning

Firewire

Firewire 800

A good GPU with a good amount of VRAM

Slot-load optical drive

Instant-wake when opening the lid; instant-sleep on close

The great power brick with wrap hooks and lighted ring

Etc.

(Of course, the OS is THE killer "feature" I can't live without.)

[Boondoggle]

you forgot bluetooth buit-in.

[jcgerm]

Originally posted by Link:
Because the compaq doesn't perform better.

You yourself might have pointed this out. While apple typically uses the most powerful chipset they can get for their processors, that compaq will have whatever they chose to give it the biggest profit margin (except in this case it's limited to intel chipsets.. hrm.. maybe the rest of the board is crap?)

Either way I wouldn't reccomend anything less then a thinkpad or alienware machine (I'd have to look at their motherboard though) in comparison to the powerbook: THE GOOD centrino notebooks tend to be equal to or more expensive then powerbooks.

OS X uses the OpenGL graphic subsystem, MS uses DirectX/Directdraw for their graphic subsystem, which has 2 speeds: slow and stable, and fast and crappy. OpenGL is slowly becoming the standard (and even used in longhorn if I'm correct) as it's cross-platform compatible with pretty much any OS.

While the PPCG4 is pretty much now in a ditch performance wise I wouldn't worry: The g4 was not IBM's fault, the g5 was and that is a rather nice processor. The portable versions of this CPU are still off in the future but I'm sure apple has plans whatever they may be.

Actually DirectX is supposed to be in Longhorn, and DirectDraw isn't a separate entity anymore. It's one of the many subsets of DirextX. Plus DirectX isn't just a graphics API. It include Direct3D, DirectPlay, DirectInput, and DirectSound among others. Both API's (DirectX and OpenGL) have their advantages, but at the rate that DirectX is progressing, OpenGL is going to need some major updating in the future to compete.

[spatterson]

Originally posted by Boondoggle:
you forgot bluetooth buit-in.

No nagromme mentioned it...

Three kinds of wireless (a, g, BT)

BT=Bluetooth

however I believe its wireless B and not A.

[CPU]

Either way I wouldn't reccomend anything less then a thinkpad or alienware machine (I'd have to look at their motherboard though) in comparison to the powerbook: THE GOOD centrino notebooks tend to be equal to or more expensive then powerbooks.

Um no, not always. I have a really well built Compaq X1000 with a 1.4Ghz chip that was only $1300. Anyone looking for a good centrino should check the X1000's out. Compaq hasn't been known for great laptops, but after doing a lot of research i think this is the best price/performance pc laptop you can get.

That being said, It still does not run osx and, most likely, if you are reading this forum, osx is the os that you prefer. So comparing pc's and macs is a moot point.

Both "sides" (pc and mac) have their respective advantages and disadvantages. As long as you are happy with your computer, who cares what the other guy has?

cpu

[The Placid Casual]

Yaawwwnnn, this subject yet again....

Need I say it?... PCs, do not run OS X. (Officially at least.)

I rest my case.

[vinster]

Originally posted by nagromme:
When price comparing to a PowerBook, don't just look at the "first glance" features like MHz and HD size. PowerBooks offer a LOT of extra functionality that's not present on many Windows laptops:

<snip>

A good GPU with a good amount of VRAM

Good point. A colleague of mine just bought a new-model Sony Vaio and it came with only 16MB of VRAM.

[Scooterboy]

I think the major drawback of the PowerBook currently is it's slow (133 - 167 MHz) system bus. The Motorola G4 can only accomodate these slow bus speeds, and has a maximum level 2 cache of 512KB. The Pentium-M system bus frequency is 400 MHz. It also has a 1MB on-die level 2 cache.

This bottleneck will be changed when the IBM G5 goes into the PowerBooks. G5's system bus is 1/2 processor speed, so a 1.2 GHz G5 would have a 600 MHz bus speed.

[RooneyX]

Originally posted by vinster:
Good point. A colleague of mine just bought a new-model Sony Vaio and it came with only 16MB of VRAM.

I saw an Alienware laptop with an upgradable 128MB GPU and as fast as leading desktop CPU.

Why use a crappy Sony as an example?

Personally, I only use Powerbooks, but I wouldn't complain if I had one of those awesome Pentium-M chips and upgradable graphics.

[vinster]

Originally posted by RooneyX:
I saw an Alienware laptop with an upgradable 128MB GPU and as fast as leading desktop CPU.

Why use a crappy Sony as an example?

Because Alienware are a bad example of a laptop. They're heavy, hot and have terrible battery life. Sony Vaios, on the other hand, are probably the closest to PowerBooks in terms of style and features.

[Dr.Michael]

I own a Thinkpad T40/1500MHz and a Powerbook 12/1000MHz.

The Thinkpad runs XP and Suse 8.3 and in most of the cases its faster than my Powerbook.
In most:
Highly G4 optimized applications are an exception. If I encode a dvd into SVCD the powerbook makes 10.4 frames/sec, the Thinkpad only 7.5.

If I run my Java programs that really need power (big decimal number crunching), the thinkpad is nearly twice as fast as the powerbook.

But Apple has still problems with the java implementation. On Panter everything became much faster in terms of java (factor 20 or more).

But: there is no day in my life that I don't have to swear and do things that my blood pressure does not appreciate when I work with the thinkpad. Don't get me wrong, the hardware quality is excellent. But the hardware-software integration is terrible in terms of linux and only ok in terms of Windows.

Compared to that, the powerbook is a pleasure to work with.

So make a decision:
If you like to tinker around with non working programs and os-features go for a thinkpad or a hp or whatever.

If you want to waste your money on poorly build machines, buy a Dell or a supermarket notebook.

If you have a clear idea what you want to do with your computer and don't want any suse, ibm or M$ related bugs or shortcomings come into your way, keep your fingers away from any X86 machine.

Your life will simply be more relaxed and you will have time to be creative (instead of looking for configuration files and entries).

[runejoha]

My original question was: Can anybody tell me i am wrong about the big gapp between the Intel Pentium M and the G4?

Nobody could, I only got this answer which was related:

"I think the major drawback of the PowerBook currently is it's slow (133 - 167 MHz) system bus. The Motorola G4 can only accomodate these slow bus speeds, and has a maximum level 2 cache of 512KB. The Pentium-M system bus frequency is 400 MHz. It also has a 1MB on-die level 2 cache."

Ah, so it is not ONLY the processor which is slower, ALSO the system buss!! This is very bad, I hope apple release the PB G5 pretty fast.

I do understand many of the answers, even though they migh not be very technical. A PB has a very good OS and it is less stress to configure and use, but is this the only advantage? Do you have to accept that the x86 systems today have better performance?

[chabig]

My original question was: Can anybody tell me i am wrong about the big gapp between the Intel Pentium M and the G4?

Where did you ask that question? I don't see it in your original post.

But of course you're not wrong. The Pentium M is 0.37GHz faster than the G4. Are you happy?

Now remember...the G4 is 3 or 4 years old, while the Pentium M was released just last year. So it's not surprising that the G4 is slower.

For the past few years, Apple notebooks have kicked butt over Wintel notebooks. Now the lead has shifted. So what. We'll leapfrog again.

Chris

[PowerMatt]

Then buy a PC laptop.

[anaphora68]

I just wanted to throw my opinion into the fray on this. If you're an apple user, what difference does it make how much faster an Intel Pentium M is? As long as my Apple does what I need it to do in a way that is fast enough, I really could care less what Intel has been pushing out the door.

Apple will catch up, that's for sure.

[runejoha]

So, the processor and the systembuss is slower than a Dell which costs the half price, if we only look at the performance.

Of course I know that Apple machines have other advantages, but because they have monopoly in manufacturing the computer and the OS, this effect could be a result of this. As we know, monopoly is bad for both the customers and the company itselfs, or at least its products.

Intel and AMN are so good today mostly of the competition they give eachother, the bad excuse for system performance a G4 has today is rediculous in a technical way of speaking.

So, why spend $500 more on a "old" Powerbook vs. a Compaq:

- Better OS (if you do not prefer to run Linux or BSD )
- Better design
- The fact that the users are used to these products.

I do want to point out that both the sytembuss and the CPU no longer are the bottle necks, and that spending money in other areas could be a better strategy for Apple! But this thread was only ment for get some constructive feedback about The CPU.

[Scooterboy]

This is a good thread, runejoha. Apple is really lagging at the moment in PowerBook performance. It's not entirely Apple's fault beyond their reliance on Motorola as the sole provider of their Pro mobile CPU. That will change as IBM CPUs work their way into the product line.

Apple has done much with what is an inferior CPU (compared to Pentium-M) and kept the PowerBooks competitive with PC laptops. They've done this with superior features and design. But a pretty case and all the ports in the world won't help if PowerBooks stay on Motorla G4's with slow system bus and infrequent releases of faster processors.

Why do I lust for a PowerBook? Is it the G4? I can get an iBook G4. Is it for sheer computing speed? There are PC laptops that are much faster and some with better graphics chipsets.
I lust for the design, the ports (especially FW800 and DVI out) and the Radeon 9600, but all that isn't enough to justify the prices if I can get an iBook that is almost as fast for most uses, has an almost as good graphics chip, and only lacks a few features compared to PowerBook, or a PC is much faster if that is what the user needs/wants.

G4 currently cannot compete with Pentium M. Not in performance or energy draw, two areas where G4 used to surpass Pentium. Right now , Pentium M is the top laptop CPU. IBM has the CPU Apple needs, and it won't be too much longer before the PowerBook G5 (sometime next year).

[squish]

Windows doesn't have Quartz Extreme. boom.

So when you drag a window around, it drags a "latent redraw trail" on all windows underneath it. I just saw it yesterday on my friend's "mobile" P4 2.53 Ghz w/ Geforce4 440go (WinXP).

[Scooterboy]

Support for Quartz Extreme is enabled by the Graphics Processor, not the CPU. My iBook doesn't support QE and I run OS X Jaguar.

Pentium 4 is not Pentium M. PC notebooks with the Pentium M CPU's are part of the Intel Centrio design.

[alien]

Before I begin, let me say that I don't doubt that the Pentium M processor may offer much better performance than a current G4. (Depending on what kind of use of course.) But I have some comments on the original post and the performance speculations there.

1. There may be more RISC instructions than CISC instructions in a program compiled for two different platforms. (I haven't seen any real-world comparisons of this.) But remember that the first thing a modern x86 chip does after fetching instructions from cache, is to break them into RISC-like instructions. The functional units in both RISC and modern CISC processors will basically do the same kind of work on the same kind of instructions. I can't see that a G4 should have any disadvantage here.

2. Yes, the Pentium M has higher clock speed than the G4. But one of the primary goals with Pentium M was to offer more performance per clock compared with the P4. Clock speed in itself won't tell that much wrt performance.

3. I think you're taking a wrong approach wrt clock cycles per instruction. The processor isn't doing one and one instruction, spending a certain number of cycles on each. It's working on many instructions at once. It's more informative to look at how many instructions the processor can complete per cycle. This will depend on how superscalar the processor is, and how many of the different types of functional units there are. As I haven't seen any discussion of the Pentium M core and how it differs from the original P4, it's hard to say how much work it can complete per cycle. But the P4 and the G4 are so different that you have to give it a thorough discussion, before coming up with clear ideas of how they perform against each other. I can only assume this is the case with the Pentium M vs. G4 as well. And if you were going to use programs that could use SIMD-style programming, that would even more complicate this issue, since AltiVec is usually a lot better than SSE2.

So even if your conclusion is right, that the Pentium M will give overall better performance than the G4, I still believe the discussion leading up to it is fundamentally flawed.

Now, let's say a Pentium M laptop indeed offers better performance along with long battery life compared with the PowerBook G4. You're suggesting Apple should change to Intel processors. But during the last years, before the Centrino/Pentium M, if we applied the same logic, everyone should have been switching to whatever technology the PowerBook was using. The PowerBook has in general been the best laptop the last couple of years. This switch didn't happen, so why bother making all this fuzz about the lack of competitive PowerBooks now?

On to the PB screen resolution, I think the basic idea is to provide a screen that most users will be happy with. With current lcd technology, many find that too many pixels crammed into a relatively small screen is hard on the eyes. I'd rather buy a cheap external lcd that I'd hook up with the PowerBook at home or the office. In fact, that's probably what Apple is calculating with. If they were to provide more screen resolution options, which would admittedly be nice, they would also have to take into account increased inventory and retail channel expenses, and lower margins. If they knew it would pay off, they'd probably do it.

The last few comments are about some programs not existing for OS X. This argument can easily be turned against Windows, as with the exception of iTunes, iApps don't exist for that platform, and the same goes for with Final Cut Pro/Express, Keynote, OmniApps, and a ton of nice Mac shareware and freeware. Not to mention all the Unix applications easily available. And if you're serious about the document workflow that many PowerBook users are into, neither ColorSync or AppleScript exist for Windows. Lastly, to me as a developer, I think it's wonderful that Apple provides their development platform for free, and I can't find the Cocoa frameworks anywhere else than under OS X.

[hldan]

I'll add one more note to this. Why is everyone going by numbers alone. I have the 17" 1.33Ghz Powerbook and it's really fast. In fact it's the fastest notebook I have seen out there.

The Pentium M is the best thing that Intel has done but if anyone wants to look at specs alone to compare speed without actually doing hands on tests then take a look at the last 2+ years how Intel tried to prove that only the megahertz numbers will show how fast the computer is and then now the Centrino is clocked at 1.3-1.7Ghz.
I can't believe nobody has acknowledged that the Centrino's only ship with 802.11b wireless. So old school. Apple has been there-done that over 2years ago with the iBook.
802.11g is the new standard for speedy wireless transmission and Apple's been all over that before anyone else.
Anyway OSX really never got good until Panther. The OS has a lot to do with speed. Mac OSX is still new and Panther has made it into primetime.
Windows XP is not new. It's just Windows 2000 dressed up with some driver enhancements so new PC's don't have to be designed to work well with it because it's nothing new to adopt to.
People that compare Macs to PCs only do that because they really want a Mac but sometimes are afraid to admit it. That's how I switched.
One last thing, the G4 processor "is" capable of 1MB L2 Cache.
The early Tibooks and the Powermacs all shipped with 1MB L2 Cache. Later Apple did cripple down to 256k L2 because they added 1MB L3.
Why the newer Macs including the G5 have only 512k L2 is either due to costs or the newer G4 processors are efficient enough to run with 512k.
I have no issues with speed using 512k L2 Cache.

[runejoha]

Alien says:
"remember that the first thing a modern x86 chip does after fetching instructions from cache, is to break them into RISC-like instructions. The functional units in both RISC and modern CISC processors will basically do the same kind of work on the same kind of instructions. I can't see that a G4 should have any disadvantage here."

Well, this is a good point. The RISC uses more CICS instructions and vice versa. A mix of these is the "best" solution. I was here thinking about a "clean" RISC processor (like MIPS etc.), but of course this is not the real picture today. BUT, if the performance, independent if it is a RISC or CISC, and we assume that they can to more or less the same number of instructions at the same time, then the cpu with fastest clock rate is the one with better performance.

"So even if your conclusion is right, that the Pentium M will give overall better performance than the G4, I still believe the discussion leading up to it is fundamentally flawed."

Maybe so, the original was not ment to be a fact but a start of a discussions to compare these processors. I have problems to invest in a powerbook becasue of these questions, which is the reason I am interested in this subject.

" But during the last years, before the Centrino/Pentium M, if we applied the same logic, everyone should have been switching to whatever technology the PowerBook was using. The PowerBook has in general been the best laptop the last couple of years. This switch didn't happen, so why bother making all this fuzz about the lack of competitive PowerBooks now? "

I do not beliveve this is true. For 2 years ago there was no problem to buy a Intel Pentium 3 1,33 MHz processor in a laptop that lasts for 4 hours(battery life) This example is the Ev0 600c Compaq. This system alone is probably better than the G4 is today, which is a reflection we should think about.

"On to the PB screen resolution, I think the basic idea is to provide a screen that most users will be happy with. With current lcd technology, many find that too many pixels crammed into a relatively small screen is hard on the eyes. I'd rather buy a cheap external lcd that I'd hook up with the PowerBook at home or the office. "

With good resolution you can use the computer in plains, trains etc. and when you are at home you no not need to hock up with another screen. I have no problems to use my screen with high resolution at all. In fact, I find it one of the most turn off areas that Apple no not support better resolution. The CPU and the system buss are not the big problem anymore anyway.

"If they were to provide more screen resolution options, which would admittedly be nice, they would also have to take into account increased inventory and retail channel expenses, and lower margins. If they knew it would pay off, they'd probably do it."

Because Apple has monopoly for macs they have calculated that bad resolution sells anyway, and there are no neighbour that are going to comepte with them. I think this is a monopoly effect.

"This argument can easily be turned against Windows, as with the exception of iTunes, iApps don't exist for that platform, and the same goes for with Final Cut Pro/Express, Keynote, OmniApps, and a ton of nice Mac shareware and freeware"

Even thoug this is true, most tings are released for windows. Many companies which are writing apps needs to prioritate windows, if this is in health care (Astra seneca, Intergagon etc.) or machine archs (Xilinx etc.) or in other areas. The Java versions for Apple are bad compare to the original Sun, the gamers have to wait for 1 year or more if they want to play a new game ETC. I am thinking of buying a mac because i allready have a PC and because I really like the OS X, but I could never survive only with mac.


"Not to mention all the Unix applications easily available."

This is a good argument!

", I think it's wonderful that Apple provides their development platform for free, and I can't find the Cocoa frameworks anywhere else than under OS X."

I didn't know! Good argument!

hldan wrote:
"802.11g is the new standard for speedy wireless transmission and Apple's been all over that before anyone else."

I know that Apple has been on the top in wireless for years.

"Anyway OSX really never got good until Panther. The OS has a lot to do with speed. Mac OSX is still new and Panther has made it into primetime. "

Maybe, anyway Panther is the best OS i ever have seen.

"Windows XP is not new. It's just Windows 2000 dressed up with some driver enhancements so new PC's don't have to be designed to work well with it because it's nothing new to adopt to. "

XP is just a good version of win95 and NT, mixed. XP is the "best" when it works. Try to innstall a wireless card into it and it install it and put it on net before you can say "cake". It is very easy to use, but I do not believe in Microsoft anymore. I believe that some Linux distributtions and MaC OS is the future for advanced users. When XP do not work it is a hell, and the OS rottens so you have to install it every 3 or 5 months. (It is also a problem for Mac OS?)

"People that compare Macs to PCs only do that because they really want a Mac but sometimes are afraid to admit "

I am very interested in Mac and are not afraid to admit it. I just want to learn more about it, and this issue is important to understand their politics about hardware. My next laptop is for sure going to be a PowerBook, but probably or hopefully with a G5 inside, because I really have not seen to much good arguments about the leak of performance for the G4.

runejoha

[hldan]

Originally posted by runejoha:

I am very interested in Mac and are not afraid to admit it. I just want to learn more about it, and this issue is important to understand their politics about hardware. My next laptop is for sure going to be a PowerBook, but probably or hopefully with a G5 inside, because I really have not seen to much good arguments about the leak of performance for the G4.

runejoha [/B]

Good for you, go for that Powerbook but the best time to get one is when you are really ready and not when a G5 model comes out. First of all no one knows when that will happen but it may happen at Macworld. Also this same arguement for Macs and PC's will go on until all computers are sharing space at the bottom of the sea with most lawyers.
There will never be a time when everyone goes Mac or goes PC only. Everyone thought that Apple had the secret weapon with OSX and Microsoft Windows was finally going to end it all but of course that didn't happen. I thought I had the worlds best PC and could NEVER be topped when I bought my Compaq 400Mhz PentiumII that played DVDs. No burning offered at the time.

Also keep in mind that most people that leave Windows and go for Mac is because of their experience and not so much curiosity. The point I am making is you have to experience the Mac experience. Try and see how responsive it is with applications. If you live near an Apple Retail Store you should drop by there and test out their computers. They usually install most of the same apps that you will find on a Windows machine like Photoshop and MS Office So go out and get one as soon as you are ready. There is never going to be a perfect time.

[runejoha]

Of course you are right. If you are going to wait until the best moment you can never buy a computer. But, i feel that my compaq which is 2 years old can offer just the same as a new Powerbook in performance, maybe even more. I have experiences with OS X and know a lot of Linux stuff and NetBSD, which both are running on a mac. But, I would whish to have a computer that runs OS X...

The problem is that I do have problems to understand Apple these days. The same processor in the economic versions as in their pro products, and this one is out of date!

I was trying to look at this case from a technical point of view, but i didn't get very clever.. Maybe I shold try at apple.com ?

"Everyone thought that Apple had the secret weapon with OSX and Microsoft Windows was finally going to end it all but of course that didn't happen."

Hmm, I think Microsoft have a big problem. The marked for servers is getting more and more over to Linux. The marked for home-work-stations/laptop is in the hands of OS X and Microsoft (and linux, but these nerds could never get agreed for a standard! which is not sufficient for common people, besides Xwindows sucks - the API for the grapical part for linux) They talk all the time about the longhorn, but I do not believe in this. Of course they will sell because of their markedpossision, but other OSes takes more and more from their marked.


Whith the G5 in laptops Apple will take over a big part of the marked, which means that OS X will get more power. I would guess that Apple sooner or later will transelate the OS for PC after some time with the G5 in the front! When this happens, The OS X turns out to be the weapon. Panther is alone a reason to buy apple today, which is probably the main reason that they care less about that they use an old cpu for all their computers. (exception power mac)

All this information is pure guessing, but I think some of this are logical.


regards

Originally posted by hldan:
Good for you, go for that Powerbook but the best time to get one is when you are really ready and not when a G5 model comes out. First of all no one knows when that will happen but it may happen at Macworld. Also this same arguement for Macs and PC's will go on until all computers are sharing space at the bottom of the sea with most lawyers.
There will never be a time when everyone goes Mac or goes PC only. Everyone thought that Apple had the secret weapon with OSX and Microsoft Windows was finally going to end it all but of course that didn't happen. I thought I had the worlds best PC and could NEVER be topped when I bought my Compaq 400Mhz PentiumII that played DVDs. No burning offered at the time.

Also keep in mind that most people that leave Windows and go for Mac is because of their experience and not so much curiosity. The point I am making is you have to experience the Mac experience. Try and see how responsive it is with applications. If you live near an Apple Retail Store you should drop by there and test out their computers. They usually install most of the same apps that you will find on a Windows machine like Photoshop and MS Office So go out and get one as soon as you are ready. There is never going to be a perfect time.

[Phanguye]

Originally posted by runejoha:
The problem is that I do have problems to understand Apple these days. The same processor in the economic versions as in their pro products, and this one is out of date!

[snipped for length]

Panther is alone a reason to buy apple today, which is probably the main reason that they care less about that they use an old cpu for all their computers. (exception power mac)

All this information is pure guessing, but I think some of this are logical.

I think they care very much that they still have the g4 in the powerbook, I would imagine that they are pouring a huge amount of r&d into getting that chip into a svelt 1 inch thick encasing

there is no doubt that the centrino chip set is the best thing intel has done in a while, but they will just be behind again when we get a mobile version of the g5

[Link]

Originally posted by OMGWTFBBQ:
As for Alienware - all they do is brand the Sager laptops with their own shells and logos. They don't do anything special to the hardware other than perhaps make up a buzzword so that they can charge more.
I used to love Alienware - then I bought one of their systems, had massive amounts of problems, and I will now never buy from them again, nor will any company I own or consult for, nor will I recommend them to anyone.
That said - Sager makes solid laptops - they are what nearly all of the HP/Compaq laptops are really underneath the graphic branding.

Turns out a month or two ago they actually ditched the sager rebranding and have someone else do it now == smaller, lighter, and a bit better... BUT

they are still kinda ugly.

Funny you can upgrade the video card in them though hehe.

[Pierre B.]

About a G5 powerbook, all I have to say is don't hold your breath on a mobile version of the current G5, even if it goes to 90 nm. It is said these days that the 90 nm version of the G5 is ready to go to production and will be available to Apple in about 3 months from now. But it is very uncertain if the power requirements will allow its use in mobile computing. There also are indications that Motorola may produce next year an enhanced version of G4 (dual core and/or some hybrid between the 74xx and 85xx series), but who counts today on Motorola's chip manufacturing abilities? The other version of the story, very interesting at that, says that, sooner or later, Apple will change completely the processor architecture going for something like a PPC based cell. You can read the long discussion here. One thing is sure: next year will be very interesting with respect to powerbook updates since, given the stagnation or perhaps the death of the current G4 (meaning I don't see it to scale further), there is not a good candidate in the horizon. And Apple cannot leave the whole powerbook line without a decent update for a whole year or more, at the moment that the competition has already the edge and is about to release the first 64-bit (Athlon based) notebook.

Of course, there is the other face of the coin: it is time that the sad days of the G4 vs. P4 desktop debate come back, in the notebook front now.

[fraeone]

I can't speak for Pentium M but I just traded in a 1.8Ghz Pentium 4 Laptop for a G4 1ghz 12". I was afraid that it would be too slow for me, but the G4 is faster--which is a completely subjective comparison, but that's what counts anyway.

[daftpig]

Originally posted by runejoha:
XP is the "best" when it works. Try to innstall a wireless card into it and it install it and put it on net before you can say "cake". It is very easy to use, but I do not believe in Microsoft anymore. I believe that some Linux distributtions and MaC OS is the future for advanced users. When XP do not work it is a hell, and the OS rottens so you have to install it every 3 or 5 months. (It is also a problem for Mac OS?)

The main selling point for Macs is that "it just works".

[Chemmy]

Originally posted by runejoha:
But, i feel that my compaq which is 2 years old can offer just the same as a new Powerbook in performance, maybe even more.

You have to be either a troll, or incredibly foolish.

[runejoha]

Originally posted by Chemmy:
You have to be either a troll, or incredibly foolish.

No, the Pentium III is 1,33 MHz. Vs the PowerBooks' 1,0 MHz today
The Compaq Evo 600c har a systembus

The PowerBook has 133 - 167 MHz systembus which is slow....

[Chemmy]

Originally posted by runejoha:
No, the Pentium III is 1,33 MHz. Vs the PowerBooks' 1,0 MHz today

You have to be either a troll, or incredibly foolish.

[runejoha]

Originally posted by Chemmy:
You have to be either a troll, or incredibly foolish.

I suppose you presents the technical staff around here. Unpleasant question are trolls. "There are no other god than mac, you should not have other gods than me etc. Do not trust the trolls talking about Intel. Intel is the devil, does not exists!"

right..

[Chemmy]

Your "argument" is a misguided at best.

First, a 1.0Ghz G4 is faster than a 1.33Ghz PIII.

Second, PowerBooks are available with 1.33Ghz G4s.

Third, your two year old Compaq isn't comparable at all to a new PowerBook. If you had ever used one, you'd know.

[ngrundy]

What makes a processor fast:

1.The instruction count.
2.The clock cycle time
3.clock cycles per instruction

1. time that an instruction takes to execute
2. width of instruction pipeline
3. length of pipeline
4. rate at which the clock runs at
5. load store times
6. memory access latancy
7. instruction decode time.

The instruction count is larger for a RISC architecture

RISC, the R means Reduced. The I means Instruction, S Set, C Computing.

Reduced Instruction Set Computing.

A RISC (PPC/MIPS/SPARC etc) processor by definition has less instructions in comparison to a CISC processor such of those of the x86 family.

Another factor that defines a RISC CPU is that each instruction only takes one cycle of the clock to complete. This line has blurred over the last few years with the introduction of SIMD (single instruction multiple data) instructions which due to the larger payload tend to take more than one clock cycle to complete. Witness these in the form of AliVec and VMX. Usually used to increase multimedia computing.

This means that maybe Apple should change to Intel ptrocessors, at least for their laptops

This point has been raised by many over the years. Apple just committed to the IBM PPC 970, which I remind you is based on the Power 4 the CPU that IBM uses in their high end mainframe and supercomputer class systems.

Also if Apple were to switch to using x86 cpu's in their systems then they would have to support two trees of source code, one for the 'legacy' PPC and one for the 'new' x86. Though the x86 IA-32 architecture is slowly on it's march to the grave. IA-64 and x86-64 are on the way in, slowly.

Also by moving to the x86 market, Apple open them selves to 'commodity computing'. The distinct edge that they have now with a closed hardware platform is lost. The intergration between hardware and software that so many love and that only combined hardware/software vendors seem to be able to provide would be lost.

Eg, Apple - OSX/PPC, IBM os390/s390, IBM - AIX/AS400, Sun - Solaris/SPARC, SGI - IRIX/MIPS.

The Apple computers have less resolution than the PCs in average. The arguments to justify this is becasue of standardization, that is is better to read on a aplle screen etc. but is it not always to have the possibility to have good resolution?

One reson would possably be to maintain the correct aspect ratios, I'm no graphic designer or video editor so i can't comment authoritivly on this. What I can comment on is the lack of headaches I no longer experiance after moving from 1280x1024 (17" crt) to 1152x854 (17" crt) or on my laptop to 1280x854.

Bigger is not allways better, and I belive that in 15-20 years time my eyesight is going to be all that much the better for it. If you truly need 1600x1200 on a 15" LCD to be able to 'work' I highly recomend that you look at how you're working and try using the minimise function a bit more.

Another drawback is that some programs are not released on the OS X, but this is for niche endusers

While this may be a problem for a few market segments still, such as Technical Drawing who use autocad day in day out, the distance between the mac and windows platform is closing rapidly for software.

What I do want to point out is that Quantity does not mean Quality. While the mac platform may only have 2 or 3 graphics editing packages they are GOOD graphics editors, not 1 good one and 15 mediocore/crap ones. This of course applies to just about any area of software you may wish to pick.

The Apple computers are beacuse of this not a temting consume for example gamers. I even wonder if an applemachine ever can run Doom3 etc.

If I remember correctly Doom 3 was showcased publicy for the first time on a Apple G4 PowerMac. Go figure.

To add to this.

While most PC users tend to only look at the hardware platform a Mac must be looked at as a whole package. When you slap down that 3k of cash for a 17" powerbook not only do you get the hardware but you get a suite of applications that you wouldn't usually get on a PC without paying at times, thousands more.

- iApps such as iMovie, iDVD, iPhoto and iTunes
- PIM software such as iCal, iSync
- Graphics, Graphic Converter 4, Art directors toolkit
- Note taking - OmniOutliner, Stickies
- Graphing, Flowcharting - OmniGraffle
- Fax - faxSFt + inbuilt in 10.3
- font management - font book
- dvd playing software
- quick books first time mac user edition

[runejoha]

Interesting! I appreciate your thought-trough contribution!

Lets lookat your list:

>>
1. time that an instruction takes to execute
2. width of instruction pipeline
3. length of pipeline
4. rate at which the clock runs at
5. load store times
6. memory access latancy
7. instruction decode time.
<<

Comments1 to 1: This depents on which instruction is excecuted,
for example Load Word in the worst case for most RISC processors.
This mean that a process which uses LW more than other makes the
pipeline halt more, but if the controll of the cpu is good
implemented, which I expect in this case, it can put some idepentent instructions after LW in the pipeline. These factors might me a drawback for the performance.

This example shows that which kind of programm and as well compiler and software can have a major influence at the performance, whic could be a reason why some programs are very ok for apple machines bur others not. In this I guess that x86 in overall is better in average for most areas, but not very good in one. (aLl benchmark test for apple are in Photoshop...)

>>2.width of instruction pipeline<<

Which is more advanved than only that. A large width is not always good for the performance. If a hazard occure this might leed to longer recovery time. I do not know which block size G4 use, or direct mem mapping? but with a big width you have to make pretty sure that you do not get very often a page fault or a miss to often.

>>3. length of pipeline<<

This is longer for a RISC set no?, more instructions.....

>>4. rate at which the clock runs at<<

Faster for Intel Pentium...

>>
5. load store times
6. memory access latancy
7. instruction decode time.
<<


I presume that 5 and 6 would be pretty the same for to different archs like the x86 and PPC. Nr 7 is probably much faster for PPC because of the reduced instruction sett, they do neither have the instructions in memory which can happen with CISCs.

But, we maybe do forgett that the bottlenecks today are in system busses rather than in the CPUs.. The speed of the systembuss of the Powerbook do not convince.

>>Another factor that defines a RISC CPU is that each instruction only takes one cycle of the clock to complete.<<

No, this is wrong. For example some old MIPS use multi-cycles, which means that a clock cycle is LESS than to finish of one instruction each cycle. This requires more controll, muxes and buffers. I do presume that modern processors like G4 also use this because it is faster and cheaper (you save HW costs because you can reuse some state elements like the ALUs). The architecture you refer to is very old and clumsy and is not used for years. Imagine the following:

If you have a static clock single cycle this has to be just as long as the longest instruction. Imagine if this instruction takes 10 times more ns than the average instructions in time and all the time you lose. I doubt they have implemented a varriable single cycle because a controll for this is highy difficult to implement in a smart way (but I do not know if they have managed to do this).

>>Also by moving to the x86 market, Apple open them selves to 'commodity computing'. The distinct edge that they have now with a closed hardware platform is lost. <<

I think that they have to compete more and release more of their monopoly to success in the future, which means that a change to Intel or AMD is a wise ting to do (I do not believe in monopoly. People says that Mictosoft is monopoly, but at least they have to exert themselves to co-operate with the HW industry. As a matter of fact I believe that this problem is the main reason for that they are useing a CPU which is out of date! They do not have a neighbour that is stressing them. Just see on this forum! People refuse to realize that the specs for a cheap dell is better than the pro Apple! Even though the specs are just specs and that a comupter exists to perform a task for you, it is very difficult to defend todays Power Books!! Of course, another subject is all the defects on the new PB 15''!

(sorry I am just a little bit dissapointed)

>>What I do want to point out is that Quantity does not mean Qualit
<<

Agree, but quantity means compatibility which is very important in the computer world ))

>>If I remember correctly Doom 3 was showcased publicy for the first time on a Apple G4 PowerMac. Go figure.<<

Very good. Which means if Apple can release a PW G5 before I have to buy a new laptopm I probably have the possibility to play Doom3! Looking forward (Or maybe I buy a Dell with the AMD 64 cpu if they release this before the PB G5!!!)

Regards
runejoha

Originally posted by ngrundy:
What makes a processor fast:

1.The instruction count.
2.The clock cycle time
3.clock cycles per instruction

1. time that an instruction takes to execute
2. width of instruction pipeline
3. length of pipeline
4. rate at which the clock runs at
5. load store times
6. memory access latancy
7. instruction decode time.

The instruction count is larger for a RISC architecture

RISC, the R means Reduced. The I means Instruction, S Set, C Computing.

Reduced Instruction Set Computing.

A RISC (PPC/MIPS/SPARC etc) processor by definition has less instructions in comparison to a CISC processor such of those of the x86 family.

Another factor that defines a RISC CPU is that each instruction only takes one cycle of the clock to complete. This line has blurred over the last few years with the introduction of SIMD (single instruction multiple data) instructions which due to the larger payload tend to take more than one clock cycle to complete. Witness these in the form of AliVec and VMX. Usually used to increase multimedia computing.

This means that maybe Apple should change to Intel ptrocessors, at least for their laptops

This point has been raised by many over the years. Apple just committed to the IBM PPC 970, which I remind you is based on the Power 4 the CPU that IBM uses in their high end mainframe and supercomputer class systems.

Also if Apple were to switch to using x86 cpu's in their systems then they would have to support two trees of source code, one for the 'legacy' PPC and one for the 'new' x86. Though the x86 IA-32 architecture is slowly on it's march to the grave. IA-64 and x86-64 are on the way in, slowly.

Also by moving to the x86 market, Apple open them selves to 'commodity computing'. The distinct edge that they have now with a closed hardware platform is lost. The intergration between hardware and software that so many love and that only combined hardware/software vendors seem to be able to provide would be lost.

Eg, Apple - OSX/PPC, IBM os390/s390, IBM - AIX/AS400, Sun - Solaris/SPARC, SGI - IRIX/MIPS.

The Apple computers have less resolution than the PCs in average. The arguments to justify this is becasue of standardization, that is is better to read on a aplle screen etc. but is it not always to have the possibility to have good resolution?

One reson would possably be to maintain the correct aspect ratios, I'm no graphic designer or video editor so i can't comment authoritivly on this. What I can comment on is the lack of headaches I no longer experiance after moving from 1280x1024 (17" crt) to 1152x854 (17" crt) or on my laptop to 1280x854.

Bigger is not allways better, and I belive that in 15-20 years time my eyesight is going to be all that much the better for it. If you truly need 1600x1200 on a 15" LCD to be able to 'work' I highly recomend that you look at how you're working and try using the minimise function a bit more.

Another drawback is that some programs are not released on the OS X, but this is for niche endusers

While this may be a problem for a few market segments still, such as Technical Drawing who use autocad day in day out, the distance between the mac and windows platform is closing rapidly for software.

What I do want to point out is that Quantity does not mean Quality. While the mac platform may only have 2 or 3 graphics editing packages they are GOOD graphics editors, not 1 good one and 15 mediocore/crap ones. This of course applies to just about any area of software you may wish to pick.

The Apple computers are beacuse of this not a temting consume for example gamers. I even wonder if an applemachine ever can run Doom3 etc.

If I remember correctly Doom 3 was showcased publicy for the first time on a Apple G4 PowerMac. Go figure.

To add to this.

While most PC users tend to only look at the hardware platform a Mac must be looked at as a whole package. When you slap down that 3k of cash for a 17" powerbook not only do you get the hardware but you get a suite of applications that you wouldn't usually get on a PC without paying at times, thousands more.

- iApps such as iMovie, iDVD, iPhoto and iTunes
- PIM software such as iCal, iSync
- Graphics, Graphic Converter 4, Art directors toolkit
- Note taking - OmniOutliner, Stickies
- Graphing, Flowcharting - OmniGraffle
- Fax - faxSFt + inbuilt in 10.3
- font management - font book
- dvd playing software
- quick books first time mac user edition

[ngrundy]

Firstly I highly recommend you get yourself a copy of Computer Organisation & Architecture - Designing for Performance by William Stallings (6th Edition) ISBN 0-13-049307-4 (International Edition)

Comments1 to 1: This depents on which instruction is excecuted,
for example Load Word in the worst case for most RISC processors.
This mean that a process which uses LW more than other makes the
pipeline halt more, but if the controll of the cpu is good
implemented, which I expect in this case, it can put some idepentent instructions after LW in the pipeline. These factors might me a drawback for the performance.

Yes a mem[i]->cpu[reg] load instruction is one of the most expensive instructions ANY cpu can do. As Stallings notes: there are around 8 different methods to do data movement on a Pentium processor, MOV, PUSH, PUSHA, MOVSX, LEA, XLAT, IN, OUT. This does not include any arithmetric operation that may do mem[i]+cpu[reg] operations which a RISC cpu can't do.

On the Performance front within the cpu's registers a x86 cpu is at a loss. Each register is purpose pecific, so if you have to do an add of two registers you need to load the numbers you want to work with as such then say we want to multiply add1 with the addres(ult)

LOAD cpu[add1]
LOAD cpu[add2]
ADD[add1+add2 addres]
MOV [add1 multi1]
MOV [addres multi2]
MULTI [multi1 multi2 multires]

Noting that LOAD, ADD, MOV and MULTI may not actually be the real instruction names.

A RISC cpu has all of it's registers as generic use registers, so the same add and multiply would go as

LOAD[r1]
LOAD[r2]
ADD[r1 r2 r3]
MULTI[r1 r3 r4]

None of the results or data to be operated had to be moved to allow for the new arithmitic style.


>>2.width of instruction pipeline<<

Which is more advanved than only that. A large width is not always good for the performance. If a hazard occure this might leed to longer recovery time. I do not know which block size G4 use, or direct mem mapping? but with a big width you have to make pretty sure that you do not get very often a page fault or a miss to often.


RISC cpu's by design do not use indirect memory access they are all direct memory access systems. This cuts down on LOAD instruction times as it will go

LOAD[mem offset]

where an x86 system while having direct memory access will also have indirect and at times double indirection.

LOAD[mem offset] ->return new memory location
LOAD[mem offset] -> return mem location or memory data
LOAD[mem offset] -> return data

Expensive.

Branch prediction misses are more of a problem in a long pipeline like the x86 pipeline, Pipeline width refers to how many bits WIDE an instruction+data is, usually 128 or 256. The width of a pipeline refers to how much information a cpu can hold on one stage of the pipeline, the more it can hold in a single stage the more data it is operating on.

This is longer for a RISC set no?, more instructions.....

No and No. As I stated before a RISC computer is a Reduced Istruction Set Computer

x86 is a CISC or Complex Instruction Set Computer. More instructions, more silicon to contain the logic.

A G4 pipeline length is 7 steps from load to save.

A Pentium 3 was 15 or 18 steps from load to save

A Pentium 4 is 20 steps from load to save.


Faster for Intel Pentium...
This is very true, but I would like to note that an x86 must spend more time 'doing stuff' to get an instruction ready and processed. IE turning a CISC instruction into smaller uops so that they can be processed by the more RISClike backend that now inhabits x86 cpus. So they need the higher clock frequency.


I presume that 5 and 6 would be pretty the same for to different archs like the x86 and PPC. Nr 7 is probably much faster for PPC because of the reduced instruction sett, they do neither have the instructions in memory which can happen with CISCs.

On a RISC architecture load and store times are generally as quick as each other. On a CISC architecture load times can be dramaticly slower than store times due to the previously mentioned indirect and double indirect load methods.

But, we maybe do forgett that the bottlenecks today are in system busses rather than in the CPUs.. The speed of the systembuss of the Powerbook do not convince.

On laptops I agree on this, the bus is slower. On G5 Desktops Apple has the lead. Not only does the system bus start at 800mhz (the fastest x86 bus out there) and then also do 900mhz and 1Ghz, In a dual CPU configuration EACH cpu has it's own dedicated system pipeline to memory running at Front side bus speed. An IA-32 system shares it's bus in the case of the intel GTL+ bus. AMD has a simmiler system on it's 32bit cpu's it's 64 bit ones use the same hypertransport bus as the PPC970.

>>Another factor that defines a RISC CPU is that each instruction only takes one cycle of the clock to complete.<<

No, this is wrong. For example some old MIPS use multi-cycles, which means that a clock cycle is LESS than to finish of one instruction each cycle. This requires more controll, muxes and buffers. I do presume that modern processors like G4 also use this because it is faster and cheaper (you save HW costs because you can reuse some state elements like the ALUs). The architecture you refer to is very old and clumsy and is not used for years.

Actually you're absolutly wrong there. I refer you to table 13.8 on page 481 of Stallings.

a MIPS R2000 (an old mips cpu) has the following caracteristics
1 instruction size
1 addressing mode
no indirect addressing
load store is not combined with aritmatic
unaligned addressing is not allowed
1 MMU use

this CPU falls right down the line when it comes to how a RISC CPU is categorised and designed.

an 80486 has the following characteristics
12 instruction sizes
15 addressing modes
does not allow indirect addressing (b)
load store is combined with arithmitic
4 MMU uses

A note that is included by stallings with regards to the 80486
b - CISC that does not conform to this characteristic.

On later 586 and 686 chips the indirect addressing characteristic was modified to allow for it, falling in line with CISC design pattens

If you have a static clock single cycle this has to be just as long as the longest instruction. Imagine if this instruction takes 10 times more ns than the average instructions in time and all the time you lose. I doubt they have implemented a varriable single cycle because a controll for this is highy difficult to implement in a smart way (but I do not know if they have managed to do this).

All computer clocks are 'static' the clock is the syncronisation between every part of the computer. Reacent introductions of technology such as DDR, QRD and what not require this clock rate to be set in stone as they must put data onto the bus on both the rise and fall of the square wave. if this gets out of sync the system goes down the drain.


People refuse to realize that the specs for a cheap dell is better than the pro Apple! Even though the specs are just specs and that a comupter exists to perform a task for you, it is very difficult to defend todays Power Books!! Of course, another subject is all the defects on the new PB 15''!

Wireless inbuilt in 1998, DVD-R standard or an option in almost every model. Now moving to 802.11g, 2GB of ram in 15 and 17" albooks, backlight keyboard, gigabit ethernet standard, firewire 800. Where are the 'poor specs'. Wireless as an inbuilt standard only really came along in 2002 for the x86 market. It can be debated that the cpu is slower of faster until hell freezes over. but again look at the overall package, metal casing, not flimsy plastic. Oh and don't get on a high horse over the defects in the 15" powerbook, every laptop has defects. Infact my university shipped 30 centrino c500's back to dell 3 weeks after arrival out of a batch of 40. Why you ask? Dead screens and faulty trackpads. All of them within 3 weeks of arrival.

Agree, but quantity means compatibility which is very important in the computer world ))

Eh? Quantity means compatabilty? Where in hells bells did that come from?

Apple virtually bends over backwards to maintain compatability with windows and unix these days, Samba 3, CUPS, *standards compliance*, cross platform software.

What gives you compatabilty is Standards Compliance. 802.1x 802.11b/g, DVD-R, CD-R, Ethernet, ZeroConf, NTP, USB, Firewire, LDAP, Kerberos, XML. All ratified standards, excepting DVD-R at this point.

While MS goes and modifies kerberos, removes AFP support from windows server...

>>If I remember correctly Doom 3 was showcased publicy for the first time on a Apple G4 PowerMac. Go figure.<<

G5's in laptops won't happen until next year, probably late next year.

~Nick
Switched 12 months ago from a dual p3 866 w/1.5gb of ram to a 1Ghz Titanium Powerbook

[pcummins]

Hello,

> No and No. As I stated before a RISC computer is a Reduced Istruction Set Computer

> x86 is a CISC or Complex Instruction Set Computer. More instructions, more silicon to contain the logic.

This is not technically 100% correct.

People tend to confuse RISC with "less instructions" and CISC with "more instructions" but it's exactly as it says - "less instruction sets" vs "more instruction sets".

RISC aims to remove redundant (and technically complicated) instruction sets with simpler to use instruction sets (ie, no more BCD arithmetic, odd string, bit packing or wierd load/store instruction sets) that are more technically uniform and easier to deal with, esp with the PowerPC simple load and store architecture.

CISC on the other hand requires more logic and resources to decode instructions, have variable instruction lengths and generally have a lot of excess or legacy baggage hanging onto the instruction set that can't be removed easily.

[runejoha]

Stalling... I had this book as pensum for years ago. A single-cycle arch is slow and is not used today. I see that you did not answer my statements about multy-cycle and other important things. Yes, some very old MIPs cpus might had single-cycle, but this model, if i remember right, was used in the Stalling book to make people understand the consept. It is not used for years! The Organisation & Architecture book is in many areas limitaded because it made for learning from a low level. it is a good book though!


The HW in the powerbook is old and is ready to be replaced! I believe the G5 PB is here before the summer!

regards
runejoha

Originally posted by ngrundy:
Firstly I highly recommend you get yourself a copy of Computer Organisation & Architecture - Designing for Performance by William Stallings (6th Edition) ISBN 0-13-049307-4 (International Edition)

Comments1 to 1: This depents on which instruction is excecuted,
for example Load Word in the worst case for most RISC processors.
This mean that a process which uses LW more than other makes the
pipeline halt more, but if the controll of the cpu is good
implemented, which I expect in this case, it can put some idepentent instructions after LW in the pipeline. These factors might me a drawback for the performance.

Yes a mem[i]->cpu[reg] load instruction is one of the most expensive instructions ANY cpu can do. As Stallings notes: there are around 8 different methods to do data movement on a Pentium processor, MOV, PUSH, PUSHA, MOVSX, LEA, XLAT, IN, OUT. This does not include any arithmetric operation that may do mem[i]+cpu[reg] operations which a RISC cpu can't do.

On the Performance front within the cpu's registers a x86 cpu is at a loss. Each register is purpose pecific, so if you have to do an add of two registers you need to load the numbers you want to work with as such then say we want to multiply add1 with the addres(ult)

LOAD cpu[add1]
LOAD cpu[add2]
ADD[add1+add2 addres]
MOV [add1 multi1]
MOV [addres multi2]
MULTI [multi1 multi2 multires]

Noting that LOAD, ADD, MOV and MULTI may not actually be the real instruction names.

A RISC cpu has all of it's registers as generic use registers, so the same add and multiply would go as

LOAD[r1]
LOAD[r2]
ADD[r1 r2 r3]
MULTI[r1 r3 r4]

None of the results or data to be operated had to be moved to allow for the new arithmitic style.


>>2.width of instruction pipeline<<

Which is more advanved than only that. A large width is not always good for the performance. If a hazard occure this might leed to longer recovery time. I do not know which block size G4 use, or direct mem mapping? but with a big width you have to make pretty sure that you do not get very often a page fault or a miss to often.


RISC cpu's by design do not use indirect memory access they are all direct memory access systems. This cuts down on LOAD instruction times as it will go

LOAD[mem offset]

where an x86 system while having direct memory access will also have indirect and at times double indirection.

LOAD[mem offset] ->return new memory location
LOAD[mem offset] -> return mem location or memory data
LOAD[mem offset] -> return data

Expensive.

Branch prediction misses are more of a problem in a long pipeline like the x86 pipeline, Pipeline width refers to how many bits WIDE an instruction+data is, usually 128 or 256. The width of a pipeline refers to how much information a cpu can hold on one stage of the pipeline, the more it can hold in a single stage the more data it is operating on.

This is longer for a RISC set no?, more instructions.....

No and No. As I stated before a RISC computer is a Reduced Istruction Set Computer

x86 is a CISC or Complex Instruction Set Computer. More instructions, more silicon to contain the logic.

A G4 pipeline length is 7 steps from load to save.

A Pentium 3 was 15 or 18 steps from load to save

A Pentium 4 is 20 steps from load to save.


Faster for Intel Pentium...
This is very true, but I would like to note that an x86 must spend more time 'doing stuff' to get an instruction ready and processed. IE turning a CISC instruction into smaller uops so that they can be processed by the more RISClike backend that now inhabits x86 cpus. So they need the higher clock frequency.


I presume that 5 and 6 would be pretty the same for to different archs like the x86 and PPC. Nr 7 is probably much faster for PPC because of the reduced instruction sett, they do neither have the instructions in memory which can happen with CISCs.

On a RISC architecture load and store times are generally as quick as each other. On a CISC architecture load times can be dramaticly slower than store times due to the previously mentioned indirect and double indirect load methods.

But, we maybe do forgett that the bottlenecks today are in system busses rather than in the CPUs.. The speed of the systembuss of the Powerbook do not convince.

On laptops I agree on this, the bus is slower. On G5 Desktops Apple has the lead. Not only does the system bus start at 800mhz (the fastest x86 bus out there) and then also do 900mhz and 1Ghz, In a dual CPU configuration EACH cpu has it's own dedicated system pipeline to memory running at Front side bus speed. An IA-32 system shares it's bus in the case of the intel GTL+ bus. AMD has a simmiler system on it's 32bit cpu's it's 64 bit ones use the same hypertransport bus as the PPC970.

>>Another factor that defines a RISC CPU is that each instruction only takes one cycle of the clock to complete.<<

No, this is wrong. For example some old MIPS use multi-cycles, which means that a clock cycle is LESS than to finish of one instruction each cycle. This requires more controll, muxes and buffers. I do presume that modern processors like G4 also use this because it is faster and cheaper (you save HW costs because you can reuse some state elements like the ALUs). The architecture you refer to is very old and clumsy and is not used for years.

Actually you're absolutly wrong there. I refer you to table 13.8 on page 481 of Stallings.

a MIPS R2000 (an old mips cpu) has the following caracteristics
1 instruction size
1 addressing mode
no indirect addressing
load store is not combined with aritmatic
unaligned addressing is not allowed
1 MMU use

this CPU falls right down the line when it comes to how a RISC CPU is categorised and designed.

an 80486 has the following characteristics
12 instruction sizes
15 addressing modes
does not allow indirect addressing (b)
load store is combined with arithmitic
4 MMU uses

A note that is included by stallings with regards to the 80486
b - CISC that does not conform to this characteristic.

On later 586 and 686 chips the indirect addressing characteristic was modified to allow for it, falling in line with CISC design pattens

If you have a static clock single cycle this has to be just as long as the longest instruction. Imagine if this instruction takes 10 times more ns than the average instructions in time and all the time you lose. I doubt they have implemented a varriable single cycle because a controll for this is highy difficult to implement in a smart way (but I do not know if they have managed to do this).

All computer clocks are 'static' the clock is the syncronisation between every part of the computer. Reacent introductions of technology such as DDR, QRD and what not require this clock rate to be set in stone as they must put data onto the bus on both the rise and fall of the square wave. if this gets out of sync the system goes down the drain.


People refuse to realize that the specs for a cheap dell is better than the pro Apple! Even though the specs are just specs and that a comupter exists to perform a task for you, it is very difficult to defend todays Power Books!! Of course, another subject is all the defects on the new PB 15''!

Wireless inbuilt in 1998, DVD-R standard or an option in almost every model. Now moving to 802.11g, 2GB of ram in 15 and 17" albooks, backlight keyboard, gigabit ethernet standard, firewire 800. Where are the 'poor specs'. Wireless as an inbuilt standard only really came along in 2002 for the x86 market. It can be debated that the cpu is slower of faster until hell freezes over. but again look at the overall package, metal casing, not flimsy plastic. Oh and don't get on a high horse over the defects in the 15" powerbook, every laptop has defects. Infact my university shipped 30 centrino c500's back to dell 3 weeks after arrival out of a batch of 40. Why you ask? Dead screens and faulty trackpads. All of them within 3 weeks of arrival.

Agree, but quantity means compatibility which is very important in the computer world ))

Eh? Quantity means compatabilty? Where in hells bells did that come from?

Apple virtually bends over backwards to maintain compatability with windows and unix these days, Samba 3, CUPS, *standards compliance*, cross platform software.

What gives you compatabilty is Standards Compliance. 802.1x 802.11b/g, DVD-R, CD-R, Ethernet, ZeroConf, NTP, USB, Firewire, LDAP, Kerberos, XML. All ratified standards, excepting DVD-R at this point.

While MS goes and modifies kerberos, removes AFP support from windows server...

>>If I remember correctly Doom 3 was showcased publicy for the first time on a Apple G4 PowerMac. Go figure.<<

G5's in laptops won't happen until next year, probably late next year.

~Nick
Switched 12 months ago from a dual p3 866 w/1.5gb of ram to a 1Ghz Titanium Powerbook

[runejoha]

Yes, this is very true. A CISC arch contains fewer instructions but they are larger and takes more time to execute! Redused instrctions means more instructions, but they are smaler and take less time to execute. A trend today is that a x86 processors use more microinstructions (like a RISC) and a RISC use more CISC. As we can see, a mix is always the best. CISC was originally designed to remove the gap between the machine level and the high level programming languages, but some scientists in the late 70ties found that a RISC arch sometimes is faster to use.

I believe in a RISC arch (even though a G4 today has a lot of "CISC" instructions) becuase I do believe in easy solutions for a complex system, but as originally stated, today I presume that the performamnce of a "RISC" and a "CISC" (in "" because both uses eachother arch) is very similar. The difference is that the Pentium M has a faster clock rate and use less power! The G4 is old HW and should be replaced ASAP!

Even though, i preciate that people are joining this discussion! I find it very interesting.

regards
runejoha

Originally posted by pcummins:
Hello,

> No and No. As I stated before a RISC computer is a Reduced Istruction Set Computer

> x86 is a CISC or Complex Instruction Set Computer. More instructions, more silicon to contain the logic.

This is not technically 100% correct.

People tend to confuse RISC with "less instructions" and CISC with "more instructions" but it's exactly as it says - "less instruction sets" vs "more instruction sets".

RISC aims to remove redundant (and technically complicated) instruction sets with simpler to use instruction sets (ie, no more BCD arithmetic, odd string, bit packing or wierd load/store instruction sets) that are more technically uniform and easier to deal with, esp with the PowerPC simple load and store architecture.

CISC on the other hand requires more logic and resources to decode instructions, have variable instruction lengths and generally have a lot of excess or legacy baggage hanging onto the instruction set that can't be removed easily.

[runejoha]

There are no other mac fans interesting in this very important dilemma and question???

I guess that "you" with his comitts that the Pentium M is a bette buy??

[DeathMan]

You seem to have a good grasp of hardware architechture, yet you've made strangley uniformed comparisons between CPUs based purely on clockspeed. This makes me believe either, there are two people posting under one account, or you are a troll trying to bait some of the more technical members into a fight.

I think ars technica would be a better playground for you. The technical macnners, typically don't like to feed the trolls.

Also, you really need to use the Mac before you can really have any kind of intelligent discussion about which computer to buy. If you've never have tried it, you can't possibly have an adequate understanding of how the computer handles. Its like discussing the deep inner workings of a car, and deciding to buy one before you even test drive it. Until you actually sit down and try it, it doesn't matter how many cubic inches of air it can displace or how many RPMs it gets up to. What if it handles like a garbage truck?

[runejoha]

>> yet you've made strangley uniformed comparisons between CPUs based purely on clockspeed<<

With arguments! Because the PPC more CISC instructions and opposite we can assume that their perfomance is more or less the same. If this statement is true the clock speed counts a lot.

>> This makes me believe either, there are two people posting under one account, or you are a troll trying to bait some of the more technical members into a fight.

No, I am one person and no, I am not a "troll" (What is that?). I am just very curious about why 1) people are sattisfied and want to spend $3000 on old technology. 2) Why Apple sells old technology 3) Which arguments exists for buying a PowerBook exception of the Panther Os which is very good (for clients) (Because the battery lifte time is bad, the CPU is bad etc. and Photoshop exists also for PC)

I am very curious! I like very much Panther and I think Apple is very important in the future, but I can't understand that people are protecting these machines with such bad performance vs. price.

regards
runejoha

Originally posted by DeathMan:
You seem to have a good grasp of hardware architechture, yet you've made strangley uniformed comparisons between CPUs based purely on clockspeed. This makes me believe either, there are two people posting under one account, or you are a troll trying to bait some of the more technical members into a fight.

I think ars technica would be a better playground for you. The technical macnners, typically don't like to feed the trolls.

Also, you really need to use the Mac before you can really have any kind of intelligent discussion about which computer to buy. If you've never have tried it, you can't possibly have an adequate understanding of how the computer handles. Its like discussing the deep inner workings of a car, and deciding to buy one before you even test drive it. Until you actually sit down and try it, it doesn't matter how many cubic inches of air it can displace or how many RPMs it gets up to. What if it handles like a garbage truck?